I coded the link between OSMOSE and our model a while ago and it still runs well. As far as I remember it’s really plug and play: you point to the OSMOSE directory containing all the configuration files and our model does the rest. As far as I can tell our model reads all but fishing mortality rates. However if the WFS instantiation of OSMOSE made some modifications to the original source code material then we’ll have to plug those modifications back on our side as well.
Here’s a sample of how the model looks like after reading in the data. Notice how the model treat unknown deep sea far off the coast as “land”. This comes straight from the OSMOSE configuration but we can change it for cosmetic reasons later. The fish is dispersed into schools and the heatmap shows the spatial distribution of biomass resulting from that.
Also, since the map is fairly small the model runs relatively quick. Here’s an example with 100 boats targeting shrimp.
The statistics of the boats are all made up so that in this example after a few years they kill off all the shrimp (bad).
But hey, at least we know the models are interacting!
While it runs, there is still plenty of stuff to do.
The first fundamental task is to set up some measures by which we make sure that the link between Osmose and our model works fine. I’d like some histograms of what happens in the original WFS implementation when:
And see if our model produces the same kind of distributions.
It’d be nice to start putting some kind of realistic parameters in our model to see how off the mark it is. Particularly for boats we need:
It would also be neat to have some more info about the geography:
Finally some basic info about what policies are currently in place.
We need to decide on a few things for our model but in particular we need to choose whether to focus on just catchability or also retention and selectivity. Osmose takes into consideration the age of the school so it is feasible that our gear would also do so but it might be a lot of work and wonder if it is really needed.
It would also be nice to have a couple of policy questions we want to ask ahead of time so we can better predict what kind of work needs to be done.